Effects of Nestling Diet on Growth and Adult Size of Zebra Finches (Poephila Guttata )

THE AUK A QUARTERLY JOURNAL OF ORNITHOLOGY

VOL. 104 APRIL 1987 NO. 2

EFFECTS OF NESTLING DIET ON GROWTH AND ADULT SIZE OF ZEBRA FINCHES (POEPHILA GUTTATA )

PETER T. BOAG Departmentof Biology,Queen's University, Kingston, Ontario K7L 3N6, Canada

Al•STRACT.--Manipulationof the diet of Zebra Finch (Poephilaguttata) nestlings in the laboratoryshowed that a low-quality diet reducedgrowth ratesof nine externalmorpholog- ical characters,while a high-quality diet increasedgrowth rates.The growth of plumage characterswas least affectedby diet, while growth ratesof tarsusand masswere most affected. The treatments also produced differencesin the adult size of experimental birds, differencesnot evident in either their parentsor their own offspring.Diet quality had the strongestimpact on adult massand tarsuslength, while plumage and beak measurements were less affected. Analysis using principal componentsand characterratios showed that the shapeof experimentalbirds was affectedby the experimentaldiets, but to a minor extent comparedwith changesin overall size. Significantshape changes involved ratiosbetween fast- and slow-growingcharacters. The ratios of charactersthat grow at similar, slow rates (e.g. beak shape) were not affected by the diets. Environmental sourcesof morphological variation should not be neglectedin studiesof phenotypicvariation in birds. Received5 June 1986, accepted30 October1986.

MORPHOLOGICAL differences between indi- fitness, and weather was seen in the nonran- vidual birds are often assignedfunctional sig- dom survival of House Sparrows collected by nificance, whether those individuals are of dif- Hiram Bumpus following a winter storm ferent species,different sexes,or different-size (O'Donald 1973, Fleischer and Johnston 1982). members of the same sex (Hamilton 1961, Se- Recently, investigators have tried to dem- lander 1966,Clark 1979,James 1982). In grani- onstrate a genetic basis for intraspecific mor- vores such as Darwin's finches (Geospizaspp.), phological variation in birds. Several studies morphologicaldifferences have been related to have shown that variation in morphological variation in feeding efficiency at all three of traits of ecologicalinterest is heritable (see re- these levels. Differences in feeding efficiency view by Boag and van Noordwijk 1986). Al- have in turn been correlated with variation in though such analyses are difficult to perform fitness due to spatial or temporal variation in and interpret, they have provided one of the seed availability (Boag and Grant 1981, Schlu- few means available to test the plausibility of ter and Grant 1984). Intraspecific morphologi- the hypothesisthat natural selectionis the pri- cal variation in several species,such as House mary determinant of the array of phenotypes Sparrows (Passerdomesticus; Johnston and Se- maintained within avian populations. lander 1971), Red-winged Blackbirds(Agelaius One common misunderstandingof heritabil- phoeniceus;Power 1969),Great Tits (Parusmajor; ity studies is that even if variation in a trait Hamilton 1961), and others (James 1970) has withintwo different populationsis highly her- been correlated with clinal variation in cli- itable, this does not guarantee that the differ- mate. A classicexample of the correlation be- ence in the characterbetween the populations tween intraspecific morphological variation, has a genetic basis.For instance,regressions of

155 The Auk 104: 155-166.April 1987 156 PETERJ. BOAG [Auk, Vol. 104 offspring bill size on parental bill size were I report a simple experimentwith laboratory- identical and significant in two years of data raised Zebra Finches (Poephilaguttata). I show collectedfor Geospizafortis (Boag 1983). This in- that qualitative differencesin the diets fed by dicated that the variation was highly heritable parentsto their young producesubstantial dif- in both years. However, y-intercepts in these ferences in the growth rates of several mor- sameregressions differed significantly(see Boag phological characters.When these young are 1983:fig. 1). Boag(1983) hypothesizedthat dif- followed until adulthood, the different nest- ferent population densities and territory sizes ling diets produceadult birds with significantin the two yearsled to differencesin the quan- ly different morphologies. tity or quality of food fed to young, with the result that young grew to different adult sizes. Although it hasbeen known for sometime that METHODS nestling diets can alter growth rates (Bryant Adult Zebra Finches were selected at random from 1978), few authors have suggestedthat varia- an outbred laboratorycolony and mated to produce tion in growth rate contributes to adult mor- 14 pairs. Eachpair was placed in a 50 x 30 x 40 cm phologicalvariation in wild birds (but seeSmith breeding cage, provided with a wicker nest cup, and Zach 1979; James 1982, 1983; Boag 1983; shreddedburlap nestingmaterial, cuttlebone,an oys- Alatalo and Lundberg 1986). Laboratory data ter shell-grit mixture, and water. Cages were main- on birds (Allee and Lutherman 1940, Johnson tained indoors on a 14L:10D lighting schedule at 1971) and mammals (Harrison 1959, Lister and constanttemperature and humidity, visually but not McCance 1965) suggest that diet in early life acousticallyisolated from each other. Cageswere as- can affect adult size. signedat randomto one of three experimentalgroups, This possibility is supported by elegant ex- which were treated identically exceptfor the quality of food available from the end of incubation until periments on Red-winged Blackbirds (James the young were approximately14 weeksold. Parents 1983). Reciprocal transplants of Red-wings be- were removed when young were about 35 days of tween either end of two separate geographic age. Breeding under all treatments commencedsi- clines in body size and shape showed that the multaneously. transferredyoung grew to resembletheir foster Families in the "standard-quality diet" treatment parents more than their biological parents. groupwere fed finch seedmix (red and white millet, Jamesconcluded that Red-wing clines in mor- with canary, rape, and flax seed) and standardnest- phology had a large nongenetic component, ling food mixture (see Table 1) ad libitum.The seed and that there probably existed complex co- mix contained about 14% crude protein, while the variation between genetic and environmental nestling mix had about 24% protein, basedon Kjel- causesof individual differences in morpholo- dahl analyses(Skoog and West 1969). In addition, thesefamilies received minced, hard-boiled hen's egg gy. with added multivitamin-mineral powder (Perve- The heritability studies discussedabove, as line•, 8 in 1 ProductsInc., about 1 g/100 g of egg) well as the transplantexperiments (James 1983), oncea week (about 1 g wet massegg mixture for each signal a growing need to considerthe origins adult or young in cage). Three days after the egg of phenotypic variation in avian populations. ration, each cageof birds receivedapproximately 5 g Such awarenesscame early to ecologistswork- wet massof fresh lettuce or spinach.The four stan- ing with other taxa. For instance,the morpho- dard-dietpairs each produced one brood,with a total logical plasticity of plants routinely forcedbot- of 6 male and 5 female offspring surviving to 1 yr of aniststo dissectecotypic variation in their study age or more. populations before invoking adaptive argu- Families in the "high-quality diet" group were given the standardseed mixture, and a high-protein ments concerning individual differences.This nestling mixture (approximately38% protein; seeTa- approachwas epitomized by the classicecotype ble 1) ad libitum.They receivedabout 1 g per bird of transfers conducted by Clausen et al. (1940). the egg mixture every day and 5 g of fresh greens Until recently, most avian ecologistshave as- every other day. Five high-diet pairs eachproduced sumedimplicitly that birds have essentiallyde- one brood, for a total of 10 male and 4 female off- terminate growth, and that phenotypicdiffer- spring at 1 yr of age. Familiesin the "low-quality diet" group were giv- encesbetween adults accuratelyreflect genetic en the standardseed mixture, and a low-protein nest- differences, which are amenable to evolution ling mixture (about 7% protein; see Table 1) ad libi- by •,atural selection. tum. From the end of incubation on, they received April1987] ZebraFinch Morphology 157 about 5 g of greens once a week, but no egg. Five TABLE1. Composition(% dry mass)of nestlingmix- low-diet pairsproduced one broodeach, with 8 male turesused in experimentaldiets. and 8 female offspringat 1 yr. All three nestling food mixtures had roughly equal energy content [about Diet quality 3,000kcal/kg or 12.6 x 106J/kg, basedon published Start- valuesfor componentingredients; metabolizable en- Low dard High ergy content (Fisher 1972) was not measured]. Rolled oats 34.0 Nine external morphologicalcharacters were mea- Pablum(baby cereal) 24.0 suredon the original parentsbefore the experiment, Wheatgerm 10.0 on the experimentalbirds as they grew (oncea day Skim milk powder 20.0 12.0 from hatch until day 14, every 2 days until day 40, Gelatinpowder 10.0 7.6 once a week until day 98, and at 1 yr of age), and on Vitamin-mineralpowder a 1.0 1.0 1.0 the fully grown offspring later produced by the ex- Alfalfa, dried 1.0 1.0 1.0 perimental birds. Offspring of experimental birds Dry dog food 6.0 20.0 were produced by mating individuals with nonsibs Tryptophan 0.1 Cystine 0.1 from the sametreatment group. Methionine 0.2 The measurementstaken included: mass(MAS) in Wheatgerm 10.0 gramson an electronic balance;wing chord (WNG) Soyaflour 19.0 in millimeters using a ruler on the flattened right Dried lean beef powder 10.0 wing; retrix length (RET)in millimetersusing a ruler Casein 10.0 from the tip of the pygostyle to the furthest exten- Brewer'syeast 7.0 sion of the tail feathers; tarsus length (TRS) in mil- Wheat bran 12.0 2.0 limeters using dial calipersfrom the nuchal notch to Corn starch 68.0 Dextrose 12.0 the furthestextension of the leg with the foot flexed at a right angle; toe spread (TOE) in millimeters, a Perveline•, 8 in 1 Pet Products,contains vitamin A palmitate, des- measuredby pressingthe foot on a ruler and noting iccatedliver, brewer'syeast, wheat germ and alfalfa meals,soy flour, driedmilk, thiaminehydrochloride, riboflavin, pyridoxine hydrochlo- the distance between the rearmost extension of the ride, vitamin B•2supplement, niacin, choline dihydrogencitrate, vi- hind toe and the foremost extension of the middle tamin E supplement,calcium pantothenate, monosodium glutamate, toe (excluding claws);bill length at nares (BLN) in tribasiccalcium phosphate, potassium biphosphate, sodium chloride, millimeters from the tip to the anterior rim of the iron orthophosphate,manganese phosphate, magnesium phosphate, copperphosphate, zinc oxide, potassium iodide, cobalt phosphate. nares;bill depth at nares(BDN) in millimeters per- pendicular to the commissureat the anterior edge of the nares; lower bill width (LBW) in millimeters at differences between results in all cases. The raw data the widest expanseof the tomia; body length (BOD) areemphasized here, except as indicated in the shape in millimeters, measuredfrom the tips of the longest and environmental sensitivity analyses.Unless in- retricesto the beak tip of a bird gently flattenedalong dicated otherwise, SAS (SAS Inst. 1982) was used for a ruler, with the previouslymeasured RET and BLN statisticalprocedures, with interpretationsbased on later being subtracted.All measurementstaken dur- Sokal and Rohlf (1981). ing growth were made by one technician, while I made all adult measurements. RESULTS Growth coefficients were calculated for each character in each experimental chick from the measure- Pairs on the low, standard, and high diets mentsfrom day 1 to 98 in nonlinearregressions based had respective averages of 3.2, 2.8, and 2.8 on a logistic model of growth (Ricklefs 1984). The SAS procedure NLIN (SAS Inst. 1982) was used to young surviving to at least one year of age, estimate K (relative growth rate, as proportion of as- implying that the diets had no differentialef- ymptote/day) values, as well as A (asymptoticsize) fect on breeding success.Young producedon and t• (age at inflection point, or maximum growth the three diets appeared normal in terms of rate) values. The K values and adult measurements grossbehavior and appearance,and individu- for each morphological characterwere then exam- als from each group went on to breed success- ined in detail, using two-way ANOVA to examine fully themselves(e.g. Table 4). As is often the the independenteffects of sex and diet on growth case in Zebra Finches, there was considerable ratesand adult size.The effectof diet on morpholog- variation between pairs within treatment ical shape was examined using both ratios of characters (Mosimann and James 1979) and principal groups in the extent to which they used the componentanalyses (PCA). All variableswere eval- nestlingmixtures for feeding their young. The uated for normality before statistical analysis. Al- high-proteinmixture was lesspreferred by the thoughvirtually all analyseswere carriedout on both parentsthan the standardmixture. Some par- log-transformedand raw data,there were only trivial ents given the low-protein nestling mixture 158 PETERJ.BOAG [Auk,Vol. 104

Adults GLM) on the growth coefficients(Table 2) Highpratein showed a consistent,highly significant effect of diet on the mean growth rates of all char- E•e- actersexcept RET. There was essentiallyno sex- E•o. ual differencein growth rates,and no signifi- .E cant diet x sex interactions. A factorial multivariate analysis of variance (MANOVA; Norusis 1985) showed no interaction between i•[i•/tI] Lowprotein (x.95•C.L.) diet and sex,no significantsexual difference in growth rates (F = 1.10, df = 10,21, P = 0.41), and a highly significanteffect of diet (F = 4.96, df = 20,42, P < 0.001). As an approximatemea- Age in days sure of the effect of diet on within-group vari- Fig. 1. Growth and adult size of TRS in Zebra ation in growth rates, ! used a modified Fmax Finchesraised on low (O) vs. high (0) quality nest- test (Sokal and Rohlf 1981) on the ratio of the ling diets.Standard-diet values were intermediateto squared coefficientsof variation (CV2). Varia- low and high values, and are omitted for clarity. tion amonglow-diet young was consistentlyup Points are means -+ 95% confidence limits. K values to twice that seenamong high-diet young,with are growth ratesbased on a logisticmodel of growth. the ratio significant in 5 of the 9 characters(Ta- ble 2; see also Fig. 2). usedit to a greaterextent than did parentspro- Diet and adult size.--The experimental diets vided with the standard mix. Simultaneous had significant,permanent effects on adult size. choice tests with nonexperimental breeding The effect on TRS, a character for which there birds confirmed that adults ranked the palat- is no adult sexual size dimorphism in this ability of the mixtures as low > standard > species,is evident in Fig. 1. Similar resultswere high. The morphologicaleffects reported be- seen for other characters when the measure- low probably derive chiefly from the extra ments of 1-yr-old experimental birds were boiled egg ration in the high diet, vs. heavy comparedwith two-way ANOVA (Table 3). As use of the low-proteinmix by parentsfeeding with growth coefficients,none of the diet x young in the low-diet group. sex interactions were significant. There was Diet and growthrate.--The diet composition evidenceof sexualsize dimorphismin RET and, had a cleareffect on nestlinggrowth rates,with to a lesser extent, in TOE. Of the 9 characters, the high diet causingyoung to grow signifi- 6 showed a significanteffect of diet on adult cantly fasterthan young rearedon the low diet size when sex was held constant. A factorial (Fig. 1, Table 2). Two-way ANOVAs (SASPROC MANOVA showed no diet x sex interaction, a

TABLE2. Effectsof nestlingdiet on means(2) and coefficientsof variation (CV) of relative growth rates(K) from fitted logistic curves.

Diet Low (n = 16) Standard(n = 13) High (n = 15) Test CV 2 Character • CV œ CV • CV ratio a Diet b Sex

MAS 0.190 29.8 0.255 23.4 0.299 10.4 ** *** NS WNG 0.243 9.8 0.243 6.3 0.265 4.9 * ** NS RET 0.307 8.6 0.306 8.5 0.321 6.1 NS NS NS TRS 0.276 15.0 0.316 12.3 0.370 8.1 NS *** NS TOE 0.281 18.6 0.336 15.6 0.373 12.4 NS *** NS BLN 0.124 18.7 0.150 7.1 0.148 9.6 * *** NS BDN 0.103 16.5 0.120 9.9 0.128 8.3 * *** NS LBW 0.112 18.2 0.142 12.6 0.139 20.3 NS *** * BaD 0.212 22.4 0.246 14.2 0.276 9.4 ** *** NS

F testfor largerof Low and High diet CV2 divided by smaller,with significanceindicated as *(P < 0.05),**(P < 0.01),and ***(P < 0.001). Two-way ANOVAs with no significantinteraction terms. Significance of diet and sex main effectsindicated as above. April 1987] ZebraFinch Morphology 159

TAI•LE3. Effectsof nestling diet on adult size.

Diet

Low (n = 16) Standard (n = 11) High (n = 14) Testb Character a • CV œ CV œ CV CV 2 Diet Sex

MAS 12.04 12.0 12.40 8.6 13.24 7.3 NS * NS WNG 55.96 2.5 56.25 2.0 57.12 1.2 * * NS RET 33.53 3.8 33.53 2.8 34.16 2.4 NS NS * * * TRS 13.19 6.2 13.76 2.4 14.08 3.2 NS ** NS TOE 18.69 5.9 19.15 4.9 19.96 4.6 NS ** * BLN 7.85 6.2 7.71 2.0 7.90 2.9 * * NS NS BDN 7.11 4.9 7.12 3.7 7.43 2.8 NS * NS LBW 6.71 4.5 6.75 3.1 6.91 4.3 NS NS NS BOD 65.24 3.8 65.84 2.9 68.81 2.7 NS *** NS

MAS measuredin grams, other charactersin millimeters. See Table 2 for explanation of tests. highly significant effect of sex (F = 5.96, df = moved, such componentsdefine ratios of pos- 9,27, P < 0.001), and a significanteffect of diet itively to negatively weighted characters,i.e. (F = 2.01, df = 18,54, P = 0.02). Relative vari- shape. ability of the measurementswas compared in As expected, the PCA on the growth data high- and low-diet birds as before; again, low- showeda majorPC I dominatedby the overall diet birds were more variable, though the in- trend for all characters to increase in size dur- creasewas significantin only 2 of 9 characters ing growth (Table 5). The two-way ANOVA on (Table 3). PC I scoresgave results similar to the univar- The morphological differences observed late growth rates;no diet x sex interaction was among the three groups of experimental birds evident, no sexual difference in PC I (F = 0.05, at adulthood were not observedin any of the df = 1,30,P = 0.82)was seen, but a strongeffect nine charactersin either their parents or their of diet on the growth rate PC I scores(F = own offspring (typical results shown in Table 12.61,df = 2,30, P < 0.0001)existed. PC II rep- 4 for tarsus). resenteda ratio betweenfast-growing leg mea- Diet and shape:principal components.--I exam- surementsand slower-growing beak measure- ined the effect of the experimental diets on ments, or measurements such as RET that were shapein two ways. First I extractedprincipal not affected by the experimental diets (Table components from both the correlation matrix 5). The two-way ANOVA on PC II showed no of growth coefficientsand the correlation ma- diet x sexinteraction, a nonsignificantsex ef- trix of adult measurements. In most sets of avi- fect (F = 3.10, df = 1,30, P = 0.09), and a non- an morphometricdata, PC I is highly positively significant effect of diet (F = 0.83, df = 2,30, correlated with all the univariate variables, and P = 0.44). Only PC III showeda significantdiet canbe interpretedas "size" (Lemen 1983, Boag effect (F = 4.22, df = 2,30, P = 0.02), with nei- 1984).The secondand subsequentcomponents ther sex nor interaction terms significant.PC typically show character correlations of mixed III is not included in Table 5 as it explained signs; with much of the variation in size re- only 7% of the total variance in growth rates.

TABLE4. Adult TRSsize (in mm,œ + SE)and ANOVA for ZebraFinches raised on 3 diets,with correspond- ing datafor their parentsand offspring.The other8 charactersshowed similar patterns.

Diet

Low Standard High ANOVA Parents 14.01+ 0.13 14.23+ 0.12 14.30+ 0.17 F2,2s= 1.16NS ExperimentMs 13.19+ 0.20 13.76+ 0.10 14.08+ 0.12 Fz•s = 9.32***a Offspring 13.79+ 0.21 13.80+ 0.10 14.08+ 0.27 F2,23= 0.68NS • P < 0.001. 160 PETERJ. BOAG [Auk,Vol. 104

TABLE5. PCAs on correlation matrices for growth K Coefficients PCA rates (n = 36) and adult measurements (n = 41). Entries are correlations between PCs and univariate measuresa (see also Fig. 2).

Adult Growth rates measurements

Character PC I PC II PC I PC II

MAS 0.75 -0.07 0.64 -0.46 WNG 0.79 0.04 0.71 -0.05 RET 0.71 0.43 0.56 0.39 TRS 0.82 -0.52 0.83 -0.17 TOE 0.83 -0.46 0.76 -0.05 BLN 0.91 0.30 0.61 0.46 BDN 0.95 0.21 0.84 0.15 LBW 0.75 0.42 0.63 0.43 BOD 0.88 -0.30 0.60 -0.64 -2 -1 0 1 Eigenvalue 6.03 1.06 4.46 1.23 PC I Percentage variation 67.00 11.80 49.60 13.60 Adult Measurements PCA ' Boldfacecorrelations are not significant;all othersare significant at P < 0.01.

It representedpositive correlationswith WNG and RET, and negative correlationswith the beak measures. The PCA on adult measurements showed that PC I summarizesoverall size variation among O -1 adults, with strong, positive correlationson all characters (Table 5). As for most of the adult -2' univariate characters,PC I showsno significant sex or interaction terms, but a highly signifi- -3' cant effect of diet on multivariate "size" (Table -3 -2 -1 0 1 2 6). In this and other analyses conducted on PC I similar suites of characters for adult Zebra Finches, PC II consistentlydefines an axis of Fig. 2. Distribution of PCA scores base;:ton sexualshape dimorphism. It weighs MA$ and growth-rate variables measured for Zebra Finches BOD negatively, and RET and the beak mea- raised on low (solid line), standard (dashed line), and sures positively; male Zebra Finches have a high (dottedline) diets(top). Bivariateroearks for each group are indicated(+). Bottomgraph showssimilar smallerbody than females,but largerbeaks and results, based on a PCA of adult size measurements. longer rectrices.PC II in adults had no inter- L = low, S = standard,H = high. action term, a highly significant effect of sex, and a significant effect of diet (Table 6). The means(Table 6) indicate that the high diet ren- ability (Fig. 2). They accountfor overtwo-thirds dered the morphology of all birds, indepen- of the total phenotypic variation in growth and dent of sex, more "female-like," while the low adult data.The experimentaldiets primarily af- diet tended to make all birds more "male-like," fected overall growth rate and adult size and with smaller bodies in relation to beak mea- had smaller effects on shape. The low-diet sures and RET. PC III from adult measurements younghad morevariable growth ratesand adult explained 9.8% of the total variation, and sizes(Fig. 2). Much of the variation in pheno- showed no significant variation among scores typic responseto the experimental diets ac- due to diet, sex, or interaction terms. tually was due to between-broodvariation (Fig. Plots of PC I vs. PC II provide a useful sum- 2). Young from the samebrood tended to clus- mary of the overall effectsof the experimental ter together,with two pairs of breedingadults diets on size, shape, and interindividual vari- on the low diet producing young with growth April 1987] ZebraFinch Morphology 161

T^I•I•E6. Effectsof diet on several indices of adult shape (œ+ SE). All entries have been multiplied by 10 for display purposes.

Diet ANOVAs a Character Low (n = 16) Standard(n = 14) High (n = 11) Diet Sex PC I -5.24 + 2.94 -2.22 _+ 1.42 7.73 _+ 1.43 ** NS PC II 3.65 _+ 2.66 -1.08 _+ 2.69 -3.32 _+ 2.48 * *** MASTRS b -7.60 _+ 0.05 -7.75 _+ 0.04 -7.75 _+ 0.05 NS NS WNGTRS 6.28 _+ 0.05 6.11 _+ 0.03 6.08 + 0.04 ** NS RETTRS 4.06 _+ 0.07 3.87 + 0.04 3.85 _+ 0.04 ** * TOETRS 1.52 _+ 0.04 1.43 _+ 0.07 1.51 _+ 0.05 NS ** BLNTRS -2.25 _+ 0.05 -2.52 _+ 0.05 -2.51 _+ 0.05 *** NS BDNTRS -2.68 _+ 0.04 -2.87 _+ 0.06 -2.78 + 0.04 NS NS LBWTRS -2.93 + 0.07 -3.10 _+ 0.06 -3.09 _+ 0.06 * NS BODTRS 6.95 + 0.07 6.80 _+ 0.04 6.89 + 0.04 NS NS BLNBDN 0.43 _+ 0.06 0.35 _+ 0.06 0.27 _+ 0.03 NS NS BLNLBW 0.68 + 0.06 0.58 _+ 0.04 0.58 + 0.04 NS NS BDNLBW 0.25 + 0.04 0.23 _+ 0.04 0.32 _+ 0.04 NS NS

Two-way ANOVAs, with no significantinteraction terms. Significance of diet and sexmain effectsindicated by *(P < 0.05), **(P < 0.01), and ***(P < 0.001). MASTRS is an abbreviationfor log(MAS) - log(TRS).

and adult PC I scoresin the standard-to high- Kruskal-Wallis tests led to identical conclu- diet range. If a single charactersuch as TRS was sions. used to calculateaverage within-brood CVs, the Diet and shape:character ratios in adults.--Mos- growth rate meanswere 7.12, 4.31, and 4.93 for iraann and James (1979; see also James 1983) the low-, standard-,and high-diet young, re- suggestedthat explicit definition of size and spectively,while the correspondingvalues for shape variables is preferable to multivariate adult TRS were 2.80, 2.00, and 1.81 (given the measuresthat may not be readily comparable small number of broods, none of the ratios of between studies.They recommendedthat dif- CV2 were significantin either case).This sug- ferences between the logarithms of various geststhat although much of the variation was charactersand a single size measure (such as between broods, the residual within-brood WNG or TRS) are appropriateindices of shape, variationmay remain higher amongsibs reared insofar as they representratios on an arithme- on low comparedwith high diets. The hetero- tic scale, but take into account the fact that scedasticityamong treatment groups was not morphological data tend to be log-normally severe enough to compromise the ANOVAs distributed. describedearlier, as replicate analyseson log- I chose TRS as a measure of size in Zebra transformed data as well as nonparametric Finches. TRS does not display sexual size di-

TABLE7. Correctedexperimental group deviationsfrom grand means,in SD units. L+H is the sum of the absolutevalues of the low- and high-dietdeviations, and providesa measureof environmentalsensitivity (see text).

Growth rates Adult measurements Character Low Standard High L+H Low Standard High L+H MAS -0.81 0.38 0.79 1.60 -0.44 -0.13 0.60 1.04 WNG -0.36 -0.20 0.55 0.91 -0.34 -0.13 0.49 0.83 RET -0.23 -0.17 0.38 0.61 -0.11 -0.16 0.25 0.36 TRS -0.69 0.00 0.85 1.54 -0.65 0.17 0.61 1.26 TOE -0.58 0.24 0.58 1.16 -0.45 -0.06 0.56 1.01 BLN -0.66 0.37 0.48 1.14 0.10 -0.34 0.15 0.25 BDN - 0.76 0.18 0.70 1.46 -0.31 - 0.30 0.59 0.90 LBW -0.71 0.58 0.41 1.12 -0.22 -0.12 0.35 0.57 BOD -0.65 -0.05 0.72 1.37 -0.56 -0.31 0.88 1.44 162 PETERJ.BOAG [Auk,Vol. 104 morphism in this species,has been used by the growth data, the mean deviations for low James(1983) for the samepurpose, and, along and high diets were -0.61 and 0.61, respec- with BDN, is the charactermost highly corre- tively, while the comparablemeans for the adult lated with PC I for the adult measurement data data were -0.33 and 0.85. This suggeststhat (Table 5). The resultsof two-way ANOVAs on the two diets elevated and depressedgrowth the ratios [actually differencesbetween log- ratesby about the sameamount, while the high transformed values, e.g. the abbreviation diet increased final adult size more than the MASTRS indicates log(MAS) - log(TRS)] be- low diet diminished it. tween each character and TRS, as well as some ratios between the beak measures,are given in DISCUSSION Table 6. Once again, none of the diet x sex interactionswere significant,and only 2 of 11 The quality of food availablefor rearing Ze- ratios showed significanteffects of sex.Four of bra Finch nestlings not only affects growth 11 ratios showed significant effectsof diet, all rates,but also has permanent effectson adult reflecting the fact that TRS is more sensitiveto size. Low-quality diets increasethe variation in diet than any of the feather or beak measures. growth rate and adult size both among and None of the beak shapemeasures gave any hint within broods. The possibility that environ- of diet group differences,indicating that subtle mental effectssuch as the protein level in a diet shape changes among similar, slow-growing might contribute to spatial or temporal mor- characterswere not taking place. phological differences in birds has been con- A measureof environmentalsensitivity.--To sidered by only a few workers to date (Smith compare the effectsof the experimental diets and Zach 1979; James 1982, 1983; Boag 1983; on the nine characters for both growth rates Alatalo and Lundberg 1986). To interpret the and adult measurements, with sexual dimor- results of my experiment, it is important to phism statistically removed, a standardized consider first if the small differences created measureof sensitivity to the dietary manipu- here could be of ecologicalsignificance, and, lation was created. First, all the growth coeffi- second,whether suchdifferences are likely to cient and adult measurement data were con- be produced under natural conditions. verted to Z scores (mean of zero, standard To evaluate whether the morphological dif- deviation of one). Using the SPSS-Xprocedure ferencesproduced here could be of ecological ANOVA (Norusis 1985), two-way ANOVAs on importance, consider the difference between diet and sex were performed with MCA (mul- high- and low-diet means in Table 3 as a per- tiple classificationanalysis). Such an analysis centageof the low-diet means.The effectof the gives the deviation from the grand mean of high diet is to increasethe low-diet adult means each experimental group, in standard devia- 1-2% for BLN or WNG, or 7-10% for characters tion units, after adjusting for the other main such as TRS or MAS. Changes in means of less effects. Thus, in the case of an adult measure than 2% are difficult to detect in nature without such as TOE, diet explained 10% of the total large samplesizes, and to detecttheir function- phenotypic variation, with means for low = al impact on activities such as foraging (Schlu- -0.45, standard= -0.06, and high = 0.56 (sex ter and Grant 1984)would be equally difficult. held constant), while sex explained 9% with Effectsin the 3-10% range could be detected in means for males = 0.25 and females = -0.35 the wild, and are more likely to be of function- (effect of diet held constant). The measure of al significance.Indeed, many avian studieshave environmental sensitivity chosenwas the sum assumedthat intraspecific morphological dif- of the absolute values of the corrected high- ferencesin this range are important. and low-diet deviations from the grand mean For instance,many monogamouspassetines (L+H values in Table 7). display sexualsize differencesof lessthan 10%. There was a significant correlation between The significant differencesbetween survivors the L+H values based on the growth data and and nonsurvivors in the Bumpus House Spar- the adult data (Spearman rank correlation, n = row data (O'Donald 1973) and Isla Daphne Ma- 9, rs = 0.73, P < 0.05). Characters such as MAS jor Geospizafortis data (Boagand Grant 1981)lie and TRS showed consistently large effects, in the 2-8% range. In examples of ecotypic while LBW and RET showed small effects. In variation in avian morphology,Fretwell (1972) April1987] ZebraFinch Morphology 163 noted significant between-habitat variation in under natural conditions.In general, we do not tarsus length for Savannah Sparrows (Passer- have adequate data to answer this question culussandwichensis, 10%), Song Sparrows(Melo- confidently.A few field studieshave reported spizamelodia, 4%), and Field Sparrows(Spizella seasonalor annual variation in morphology pusilla,2%). Grant et aL (1976)found significant (Perrins 1979, Smith and Zach 1979, Boag 1983), between-habitat differences in Geospizafortis and Alatalo and Lundberg (1986) found signif- morphologyof well under 10%,while morpho- icant differencesin tarsuslength of Pied Fly- logicaldifferences of a similarmagnitude have catcher(Ficedula hypoleuca) offspring reared by been noted for titmice (Parus spp.) living in primary vs. secondaryfemales of polygynous coniferous and hardwood forests (Perrins 1979). males. The differences (about 2%) were attrib- Authors studying clinal variation in avian mor- uted to differences in paternal feeding rate. phology (James1970) or making island-main- Alatalo and Lundberg(1986) also found signif- land comparisons(e.g. island subspeciesof the icant effects of breeding density on tarsus wren Troglodytestroglodytes; see Williamson length, with nests in high-density areas pro- 1981: 139) also commonly discussthe adaptive ducing smaller offspring. Several other long- significanceof differences between popula- term studiesprobably have similar data, al- tions of well under 10%. In a recent summary though often birds have not been measuredas of temporal and geographical morphologial adults, or different workers have been in- variation in the Fox Sparrow (Passerellailiaca), volved in different areasor years,complicating Zink (1983) sought evolutionary interpreta- comparisons. tions of results for which the largest average Despite the paucity of data, there is no rea- percentagedifference was 3.28%,with a range son to believe such effects are rare in nature. of 0.55-8.47%. Adaptive explanations for sig- The experimentalregime I used does not in- nificant morphological changes of 3-5% in volve a totally unrealisticor pathologicalrange House Finches (Carpodacusmexicanus) in their of diet qualities. Many estrildid finches rou- new easternpopulations compared with their tinely raise young on seedalone in the wild ancestralwestern populations were proposed (Goodwin 1982, Morton and Davies 1983, Zann by Aldrich (1982).Dhondt et aL (1979) found and Straw 1984), and many of them inhabit sa- that Great Tits gradually became3-7% smaller vannah habitats with concomitant seasonal and over 14 years.Several mechanisms for the shift annual unpredictability in the timing, quanti- involving natural selectionwere suggested,re- ty, and quality of resourcesduring the nesting volving around the fact that the addition of season.This may result in occasional,oppor- nest boxeshad doubled or tripled Parusbreed- tunisticinsect feeding or periodic food limita- ing densitiesduring this period. tion, during which time nestlingsand fledg- In citing these studies, I am not suggesting lings starve(Zann and Straw 1984).Granivores that all such examplesinvolve morphological with faster development times than Zebra differences attributable to nonadaptive, envi- Finchesrequire a high-protein diet for breed- ronmental effectssuch as nestling diet quality. ing to be successful.Such speciesoften feed I merely point out that avian ecologistsregu- nestlingson insects(Jones and Ward 1976) or larly seekadaptive explanations for differences other protein-rich items suchas the specialized of a magnitude that I have producedby way of use of Spirogyraalgae by Sharp-tailedMunias nonpathologicalvariation in diet quality (by (Lonchurastriata; Avery 1980). Many birds de- nonpathological I mean diets not producing pendent on insects for rearing young show nestlingdeath or infertile adults).In somecases, considerable seasonal or annual variation in e.g. the trend seen by Dhondt et aL (1979) in nestling growth rates, fledging success,and Parussize over time, I see parallels with the brood-reductionbehavior (Bryant 1975, 1978; breeding-density-associatedannual variation in Perrins 1976; Howe 1977; Ricklefs and Peters offspring size noted by Boag(1983) for Geospiza 1979),implying that the potentialexists for diet- fortis. induced morphological change. Dietary pro- The second question raised above was tein can have major effectson both growth and whether variation in diet or other aspectsof adult size of captive ducks (Johnson 1971) and the nestling environment sufficientto produce domestic fowl (Fisher and Griminger 1963). significantmorphological changes are common Other environmental effects have also been in- 164 PETERJ. BOAC [Auk, Vol. 104 vestigated.For example, pullets raised at 6øC tween-broodmorphological differences that are during 3-6 months of age were shorter,gained due to environmental effects. Some effect of diet more weight, and had smaller tarsi and tails on variation within Zebra Finch broods re- than former flock mateskept at about 23øC(A1- mained after accounting for between-brood lee and Lutherman 1940). Similar studies re- variation, indicating that processessuch as sib- ported the impact of environmental effects on ling competitionmight affectintraspecific mor- morphology in mammals (Widdowson and phologicalvariation. Environmental effects that McCance 1960, Lister and McCance 1965, Wil- alter the distribution of variance within and liams and Hughes 1975, Barnett and Dickson between broods can have important conse- 1984), including humans (Hulse 1968). In one quences for heritability analyses (Falconer of the best longitudinal studiesof associations 1981).Although sucheffects are citedregularly between intraspecificmorphological variation in criticismsof heritability studiesunder field and fitness in mammals, Clutton-Brock et al. conditions, almost no data exist to support their (1982)found that diet or density-relatedmater- existence or nonexistence. Van Noordwijk nal effectscan have profound effectson body (1984)suggested that environmentaleffects re- size and the lifetime reproductive successof suiting from naturally occurringpoor breeding individuals or entire cohorts of red deer (Cervus seasonsor from manipulation of brood size can elaphus). at least temporarily increase morphological Among the Zebra Finches used in this ex- variation among sibs,and hence decreaseher- periment, there was considerablevariation in itability estimates. the responseof different breeding pairs to the The Zebra Finch data suggestthat the pri- diet treatments;one pair wasable to raiseyoung mary effect of variation in diet quality is on to "high-diet size" on the low diet. Part of this overall growth and adult size. There was not variation may be an artifact of the multicom- enough differential impact on charactersto ponent, qualitative diets used, which made it produce large changesin shape. The small difficult to control the amountsof protein, en- shapechanges observed were due primarily to ergy, or other dietary componentsconsumed differences between characters in their rate and by each young bird. Zebra Finches can raise timing of growth. Measurementssuch as MAS young on seedalone, so it is unlikely any birds or TRS increasequickly and reach asymptotes sufferedfrom an energy shortfall.Similarly, all at an early age; they are affectedmost by vari- nestling and egg mixturescontained a vitamin- ation in diet. Beakcharacters grow more slow- mineral supplement,making specificshortages ly, for a prolongedperiod. Beaks are lesssus- in these componentsunlikely. The most ob- ceptible to dietary influencesthat have their vious difference between the diets was in crude majoreffect while youngare dependent on their protein availability;however, in the absenceof parents. Charactersinvolving feather growth detailed data on assimilation efficiencies, ami- also are affected less seriously,despite rather no acid balance,or micronutrient availability high growth rates. This reflectsthe high de- (Fisher 1972), it would be misleading to con- velopmentalpriority of passerineplumage de- clude that the morphologicaleffects observed velopment, which tends to proceed according here are solely the result of dietary protein to a fixed schedule even if a nestling is starv- level. Some of these difficulties could be avoid- ing. ed if a single, synthetic food ration had been It remains to be seen if environmental effects used(e.g. Murphy and King 1982).It is possible inducedby variationin the quality or quantity to switch Zebra Finches to a synthetic diet, but of nestling diets, or other aspectsof the nest this can causeadult mortality and poor breed- environment, regularly influence natural pat- ing success,at least initially (T. Scheuhammer terns of intraspecificmorphological variation pers. comm.). The complications described in birds. Considerablescope exists for long- aboveare important in designingfuture exper- term, longitudinal studiesof the environmen- iments, but do not alter my principal finding tal correlatesof morphologicalvariation, as well that diet quality can affect adult size as well as as more natural experiments of the type I have growth rates in Zebra Finches. conductedhere (e.g. Berthold 1976, 1977;Smith Under natural conditions, differences in the and Arcese 1987). The primary data on varia- quality of parental care might produce be- tion in avian growth rateshave not been sup- April 1987] ZebraFinch Morphology 165 portedby studiesthat followed young to adult- 1977. Ober die kiinstliche Aufzucht Nest- hood to check for permanent effects on junger Amseln (Turdusmerula) reit Beeren des morphology. Hence, we cannot discriminate Efeus(Hedera helix). Vogelwarte 29: 110-113. BOAG,P. T. 1983. The heritability of external mor- which avian charactersdisplay compensatory, phology in Darwin's ground finches (Geospiza) or "target-seeking,"growth (Atchley 1984)once on Isla Daphne Major, Galapagos.Evolution 37: the period of parental dependenceis past, and 877-894. which do not. Judicious choice of characters 1984. Growth and allometry of external may well minimize the likelihood of signifi- morphologyin Darwin'sfinches (Geospiza) on Isla cant environmental effects in studies of avian Daphne Major, Galapagos.J. Zool. London 204: morphometrics.Until additional data are avail- 413-441. able, however, the possibility of environmen- , & P. R. GRANT. 1981. Intense natural selec- tal effects should not be dismissed automati- tion in a population of Darwin's finches (Geo- cally in favor of selective explanations for spizinae) in the Galapagos.Science 214: 82-85. morphological differences between bird pop- ß & A. J. VAN NOORDWIJK.1987. The quanti- ulations. tative geneticsof wild bird populations.In Avi- an genetics--its application to ecological and evolutionarytheory (F. Cookeand P. A. Buckley, ACKNOWLEDGMENTS Eds.). London, Academic Press. BRYANT,D. M. 1975. Breeding biology of House I thank Radiq Tallim, Jody Morgan, and my un- Martins Delichon urbica in relation to aerial insect dergraduateassistants for their help. L. M. Ratcliffe, abundance. Ibis 117: 180-215. J. N.M. Smith, and the refereeshelped improve the ß 1978. Environmental influences on growth manuscript.My Zebra Finch researchhas been sup- and survival of nestling House Martins Delichon ported by researchgrants from Trent and Queen's urbica. Ibis 120: 271-283. universities,as well as operating grant U-0315 from CLARK,G. A., JR. 1979. Bodyweights of birds:a re- the Natural Scienceand EngineeringResearch Coun- view. Condor 81: 193-202. cil of Canada. CLAUSEN,J., D. D. KECK, & W. M. HIESY. 1940. Ex- perimental studies on the nature of species.I. Effect of varied environments on western North LITERATURE CITED American plants. Carnegie Inst. Wash. Publ. 520. ALATALO,R. V., & A. LUNDBERG.1986. Heritability CLUTTON-BROCK,T. H., F. E. GUINNESS,& $. D. ALBON. and selectionon tarsuslength in the Pied Fly- 1982. Red deer. Behavior and ecology of two catcher (Ficedulahypoleuca). Evolution 40: 574- sexes.Edinburghß University Press. 583. DHONDT,A., R. EYCKERMAN,& J. HUBLfi. 1979. Will ALDRICH,J. W. 1982. Rapid evolution in the House Great Tits become little tits? Biol. J. Linn. Soc. 11: 289-294. Finch (Carpodacusmexicanus). J. YamashinaInst. Ornithol. 14: 179-186. FALCONER,D. $. 1981. Introduction to quantitative ALLEE,W. C., & C. Z. LUTHERMAN. 1940. An exper- genetics.London, Longman. imental study of certain effectsof temperature FISHER,H. 1972. The nutrition of birds. Pp. 431-469 on differential growth of pullets. Ecology21: 29- in Avian biology II (D. S. Farner, J. R. King, and 33. K. C. Parkes, Eds.). New Yorkß Academic Press. ATCHLEY,W. R. 1984. Ontogeny, timing of devel- , & P. GRIMINGER.1963. Aging and food re- opment, and genetic variance-covariancestruc- striction:changes in body compositionand hy- ture. Amer. Natur. 123: 519-540. droxyproline content of selectedtissues. J. Nutr. AVERY,M. L. 1980. Diet and breeding seasonality 80: 350-354. amonga population of Sharp-tailedMunias, Lon- FLEISCHER,R. C., & R. F. JOHNSTON.1982. Natural churastriata, in Malaysia. Auk 97: 160-166. selectionon body size and proportionsin House BARNETT,$. A., & R. G. DICKSON. 1984. Changes Sparrows.Nature 298: 747-749. among wild housemice (Mus rnusculus)bred for FRETWELL,$. D. 1972. Populationsin a seasonalen- ten generationsin a cold environmentßand their vironment. Princeton, New Jersey, Princeton evolutionary implications.J. Zool. London 203: Univ. Press. 163-180. GOODWIN, D. 1982. Estrildid finches of the world. BERTHOLD,P. 1976. Ober den Einfluss der Nestlings- Oxford, Oxford Univ. Press. hahrung auf die Jugendentwicklung,insbeson- GRANT,P. R., B. R. GRANT,J. N.M. SMITH,I. J. ABBOTT, dere auf das Fliigelwachstum,bei der M/Snchs- & L. K. ABBOTT.1976. Darwin's finches:popu- grasmiicke(Sylvia atricapilla). Vogelwarte 28: 257- lation variation and natural selection. Proc. Natl. 263. Acad. Sci. 73: 257-261. 166 PETERJ. BOAG [Auk, Vol. 104

HAMILTON,T. H. 1961. The adaptive significances O'DONALD,P. 1973. A further analysisof Bumpus's of intraspecifictrends of variationin wing length data:the intensity of natural selection.Evolution and body size among bird species.Evolution 15: 27: 398-404. 180-195. PERRINS,C. g. 1976. Possibleeffects of qualitative HARRISON, G. A. 1959. Environmental determina- changesin the insectdiet of avian predators.Ibis tion of the phenotype.Pp. 81-86 in Functionand 118: 580-584. taxonomic importance (A. J. Cain, Ed.). London, 1979. British tits. London, Collins. SystematicsAssoc. Publ. No. 3. POWER,D. M. 1969. Evolutionary implications of HOWE,H. F. 1977. Sex-ratioadjustment in the Com- wing and size variation in the Red-winged mon Grackle. Science 198: 744-746. Blackbirdin relation to geographicand climatic HULSE, F. S. 1968. The breakdown of isolates and factors:a multiple regressionanalysis. Syst. Zool. hybrid vigor among the Italian Swiss.Proc. 12th 18: 363-373. Intern. Congr. Genet. 2: 177. RICKLEFS,R. E. 1984. Components of variance in JAMES,F.C. 1970. Geographicsize variation in birds measurementsof nestling European Starlings and its relationship to climate. Ecology51: 365- (Sturnusvulgaris) in southeasternPennsylvania. 390. Auk 101: 319-333. 1982. The ecologicalmorphology of birds: ', & S. PETERS.1979. Intraspecificvariation in a review. Ann. Zool. Fennici 19: 265-275. the growth rate of nestling EuropeanStarlings. 1983. Environmental component of mor- Bird-Banding50: 338-348. phological differentiation in birds. Science221: SAS INSTITUTEINC. 1982. SASuser's guide: statistics. 184-186. Cary, North Carolina, SAS Inst. JOHNSON,N. F. 1971. Effectsof dietary protein on SCHLUTER,D., & P. g. GRANT. 1984. Determinants of Wood Duck growth. J. Wildl. Mgmt. 35: 798-802. morphologicalpatterns in communitiesof Dar- JOHNSTON,R. F. & R. K. SELANDER.1971. Evolution win's finches. Amer. Natur. 123: 175-196. in the HouseSparrow. II. Differentiationin North SELANDER,R. K. 1966. Sexual dimorphism and dif- American populations. Evolution 25: 1-28. ferential niche utilization in birds. Condor 68: JONES,P. J., & P. WARD. 1976. The level of reserve 113-151. protein as the proximate factor controlling the S•COOG,D. A., & D. M. W•ST. 1969. Fundamentals of timing of breeding and clutch-sizein the Red- analytical chemistry,2nd ed. New York, Holt, billed Quelea, Queleaquelea. Ibis 18: 547-574. Rinehart and Winston. LEMEN,C.A. 1983. The effectivenessof shape anal- SMITH,J. N.M., & P. ARCESE.1987. Effects of sup- ysis. Fieldiana Zool. No. 15. plemental food on growth and adult size in the LISTER,D., & R. A. MCCANCE. 1965. The effects of SongSparrow. Proc. 19th Intern. Ornithol.Congr. two diets on the growth, reproductionand ulti- in press. mate size of Guinea-pigs.Brit. J. Nutr. 19: 311- ß, & R. ZACH. 1979. Heritability of somemor- 319. phological charactersin a Song Sparrow popu- MORTON, S. R., & P. H. DAVIES. 1983. Food of the lation. Evolution 33: 460-467. Zebra Finch (Poephilaguttata), and an examina- SO•CAL,R. R., & F. J. ROHLF. 1981. Biometry. San tion of granivoryin birds of the Australianarid Francisco, W. H. Freeman. zone. Australian J. Ecol. 8.' 235-243. WIDDOWSON, E.M., & R. A. MCCANCE. 1960. Some MOSIMANN,J.E., & F. C. JAMES. 1979. New statistical effectsof acceleratinggrowth I. General somatic methodsfor allometry with applicationsto Flor- development. Proc. Roy. Soc. 152: 188-206. ida Red-winged Blackbirds.Evolution 33: 444- WILLIAMS,J.P., & P. C. HUGHES. 1975. Catch-up 459. growth in rats undernourishedfor different pe- MURPHY,M.E., & J. R. KING. 1982. Semi-synthetic riodsduring the sucklingperiod. Growth 39: 179- diets as a tool for nutritional ecology.Auk 99: 193. 165-167. WILLIAMSON,M. 1981. Island populations.Oxford, VAN NOORDWIJK,A. J. 1984. Quantitative genetics Oxford Univ. Press. in natural populations of birds illustrated with ZANN,R., & B. STRAW.1984. Feedingecology and examplesfrom the Great Tit, Parusmajor. Pp. 67- breedingof ZebraFinches in farmlandin north- 79 in Population biology and evolution (K. ern Victoria. Australian Wildl. Res. 11: 533-552. W•Shrman and V. Loeschcke, Eds.). Berlin, ZINK, R. M. 1983. Evolutionary and systematicsig- Springer-Verlag. nificanceof temporal variation in the Fox Spar- NORUSIS,M. J. 1985. SPSS-X advanced statistics row. Syst. Zool. 32: 223-238. guide. New York, McGraw-Hill.